File size: 7,852 Bytes
448903c
 
 
 
 
 
 
 
 
 
dfad45c
 
3b31030
 
448903c
 
 
 
 
 
 
 
 
 
 
dfad45c
3b31030
 
448903c
 
 
 
3b31030
 
448903c
 
 
 
 
 
 
 
 
 
 
 
 
 
69fe8b2
 
3b31030
 
448903c
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
3b31030
 
 
 
 
 
 
 
 
 
 
 
448903c
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
99bb959
448903c
 
3b31030
dfad45c
448903c
 
3b31030
 
 
448903c
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
dfad45c
448903c
dfad45c
 
 
3b31030
dfad45c
448903c
dfad45c
799013a
 
 
 
448903c
3b31030
448903c
799013a
3b31030
 
448903c
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
3b31030
 
 
 
 
 
448903c
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
import os
import requests
from dotenv import load_dotenv
from langgraph.graph import StateGraph, MessagesState, START
from langchain_huggingface import ChatHuggingFace, HuggingFaceEndpoint
from langchain_openai import ChatOpenAI
from langchain_core.messages import HumanMessage, SystemMessage
from langgraph.prebuilt import ToolNode, tools_condition
from langchain_google_genai import ChatGoogleGenerativeAI
from langfuse.langchain import CallbackHandler
from tools.web_search import web_search
from tools.math import add_numbers_in_list, check_commutativity
from tools.extraction import extract_data_from_excel, extract_transcript_from_youtube, extract_transcript_from_audio
from rate_limiters import safe_invoke_with_retry_gemini

load_dotenv(override=True)

PROVIDER="google"

langfuse_handler = CallbackHandler()

tools = [
    add_numbers_in_list,
    web_search,
    check_commutativity,
    extract_data_from_excel,
    extract_transcript_from_youtube,
    extract_transcript_from_audio
]

# --------------- Define the agent structure ---------------- #
def build_agent(provider: str = "hf"):
    USE_RATE_LIMITER = os.getenv("USE_RATE_LIMITER", "false").lower() == "true"

    print(f"Building agent with provider: {provider}")
    if provider == "hf":
        llm = HuggingFaceEndpoint(
            repo_id="mistralai/Mixtral-8x7B-Instruct-v0.1",
            task="text-generation",
            temperature=0.0,
            provider="hf-inference"
        )

        llm = ChatHuggingFace(llm=llm)

    elif provider == "google":
        # Google Gemini
        llm = ChatGoogleGenerativeAI(
            # model="gemini-2.0-flash",
            model="gemini-2.5-flash-preview-05-20",
            # model="gemini-2.5-flash-lite-preview-06-17",
            max_tokens=2048,
            max_retries=2,
        )

    elif provider == "openai":
        llm = ChatOpenAI(
            model="gpt-3.5-turbo",  # or "gpt-3.5-turbo"
            temperature=0,
            api_key=os.getenv("OPENAI_API_KEY"),
            max_tokens=512
        )
    else:
        raise ValueError(f"Unsupported provider: {provider}")
    
    # Bind the tools to the LLM
    llm_with_tools = llm.bind_tools(tools)

    # load the system prompt from the file
    with open("system_prompt.txt", "r", encoding="utf-8") as f:
        system_prompt = f.read()

    # Create system message with the system prompt
    sys_msg = SystemMessage(content=system_prompt)

    # --------------- Define nodes ---------------- #
    def assistant(state: MessagesState):
        """Node for the assistant to respond to user input."""
        if USE_RATE_LIMITER:
            if provider == "google":
                response = safe_invoke_with_retry_gemini(
                    llm_with_tools,
                    [sys_msg] + state["messages"],
                    max_retries=2,
                    wait_seconds=60
                )
            else:
                raise ValueError(f"Rate limiting is not implemented for provider {provider}.")
        else:
            response = llm_with_tools.invoke([sys_msg] + state["messages"])
        return {"messages": [response]}
    

    tool_node = ToolNode(tools=tools)

    # --------------- Build the state graph ---------------- #
    graph_builder = StateGraph(MessagesState)

    graph_builder.add_node("assistant", assistant)
    graph_builder.add_node("tools", tool_node)

    graph_builder.add_conditional_edges(
        "assistant",
        tools_condition,
    )
    graph_builder.add_edge("tools", "assistant")
    graph_builder.add_edge(START, "assistant")

    return graph_builder.compile()



# --------------- For manual testing ---------------- #
if __name__ == "__main__":
    print("\n" + "-"*30 + " Agent Starting " + "-"*30)
    # Print run variables in a table format
    print(f"Provider: {PROVIDER}")
    print(f"Search engine used: {'DDGS' if os.getenv('USE_DDGS').lower() == 'true' else 'Tavily'}")
    agent = build_agent(provider=PROVIDER)  # Change to "hf" for HuggingFace
    print("Agent built successfully.")
    print("-"*70)

    # Get questions
    DEFAULT_API_URL = "https://agents-course-unit4-scoring.hf.space"
    api_url = DEFAULT_API_URL
    questions_url = f"{api_url}/questions"
    files_url = f"{api_url}/files/" # Needs task_id

    # 2. Fetch Questions
    print(f"Fetching questions from: {questions_url}")
    try:
        response = requests.get(questions_url, timeout=15)
        response.raise_for_status()
        questions_data = response.json()
        if not questions_data:
             print("Fetched questions list is empty.")
        print(f"Fetched {len(questions_data)} questions.")
    except Exception as e:
        print(f"An unexpected error occurred fetching questions: {e}")

    # 3. Get specific question by task_id
    # task_id = "8e867cd7-cff9-4e6c-867a-ff5ddc2550be" # Sosa albums
    # task_id = "2d83110e-a098-4ebb-9987-066c06fa42d0"  # Reverse text example
    # task_id = "cca530fc-4052-43b2-b130-b30968d8aa44" # Chess image
    # task_id = "4fc2f1ae-8625-45b5-ab34-ad4433bc21f8" # Dinosaur ?
    # task_id = "6f37996b-2ac7-44b0-8e68-6d28256631b4" # Commutativity check
    # task_id = "9d191bce-651d-4746-be2d-7ef8ecadb9c2" # Youtube video 
    # task_id = "cabe07ed-9eca-40ea-8ead-410ef5e83f91" # Louvrier ?
    # task_id = "f918266a-b3e0-4914-865d-4faa564f1aef"  # Code example
    # task_id = "3f57289b-8c60-48be-bd80-01f8099ca449" # at bats ?
    task_id = "7bd855d8-463d-4ed5-93ca-5fe35145f733" # Excel file
    # task_id = "5a0c1adf-205e-4841-a666-7c3ef95def9d" # Malko competition
    # task_id = "305ac316-eef6-4446-960a-92d80d542f82" # Poland film
    # task_id = "bda648d7-d618-4883-88f4-3466eabd860e"  # Vietnamese
    # task_id = "cf106601-ab4f-4af9-b045-5295fe67b37d" # Olympics
    # task_id = "a0c07678-e491-4bbc-8f0b-07405144218f" # pitchers
    # task_id = "3cef3a44-215e-4aed-8e3b-b1e3f08063b7" # grocery list
    # task_id = "840bfca7-4f7b-481a-8794-c560c340185d" # Carolyn Collins Petersen
    # task_id = "1f975693-876d-457b-a649-393859e79bf3" # Audio (pages)
    # task_id = "99c9cc74-fdc8-46c6-8f8d-3ce2d3bfeea3" # Audio (recipe)

    # get question with task_id
    q_data = next((item for item in questions_data if item["task_id"] == task_id), None)

    content = [
        {"type": "text", "text": q_data["question"]}
    ]

    if q_data["file_name"] != "":
        file_url = f"{files_url}{task_id}"

        if q_data["file_name"].endswith((".png", ".jpg", ".jpeg")):
            content.append({"type": "image_url", "image_url": {"url": file_url}})

        elif q_data["file_name"].endswith((".py")):
            # For code files, we can just send the text content
            try:
                response = requests.get(file_url, timeout=15)
                response.raise_for_status()
                code_content = response.text

                content.append({"type": "text", "text": code_content})
            except Exception as e:
                print(f"Error fetching code file: {e}")

        elif q_data["file_name"].endswith((".xlsx", ".xls")):
            content.append({"type": "text", "text": "Excel file url: " + file_url})

        elif q_data["file_name"].endswith((".mp3", ".wav")):
            content.append({"type": "text", "text": "Audio file url: " + file_url})

        else:
            content.append({"type": "text", "text": f"File URL: {file_url} (file type not supported)"})
    
    human_msg = HumanMessage(content=content)

    human_msg.pretty_print()

    try:
        result = agent.invoke(
            {"messages": [human_msg]},
            config={"callbacks": [langfuse_handler]}
        )

        for message in result["messages"]:
            message.pretty_print()
        # Result already printed inside assistant() node
    except Exception as e:
        print(f"Error: {e}")